Punchcards for Data Archival

LETTERSHIFT FIGURESHIFT
Deci Tri Glyph Deci Tri Glyph
0000 000 <NUL> 0000 000 <NUL>
0001 001 A 0001 001 1
0002 002 B 0002 002 2
0003 010 C 0003 010 3
0004 011 D 0004 011 4
0005 012 E 0005 012 5
0006 020 F 0006 020 6
0007 021 G 0007 021 7
0008 022 H 0008 022 8
0009 100 I 0009 100 9
0010 101 J 0010 101 0
0011 102 K 0011 102 .
0012 110 L 0012 110 ,
0013 111 M 0013 111 :
0014 112 N 0014 112 +
0015 120 O 0015 120 '
0016 121 P 0016 121 -
0017 122 Q 0017 122 *
0018 200 R 0018 200 /
0019 201 S 0019 201 (
0020 202 T 0020 202 )
0021 210 V 0021 210 ^
0022 211 X 0022 211 ?
0023 212 Y 0023 212 !
0024 220 Z 0024 220 ¤
0025 221 <SP> 0025 221 <CRLF>
0026 222 <FS> 0026 222 <LS>
This is a 3-trit code I "developed". Aside from stealing the "shift" idea from Baudot Code (a 5-bit keymap), all I really had to do was count to 26 and sing the important bits of the Alphabet (U and W are vtterly vvorthless) without going full linguist and getting rid of pretty letters like C and F. "Figureshift" and "Lettershift" basically instruct the interpreter to switch to an alternate codepage, it's a nice hack to use when you realise you don't have enough numbers to do anything useful.
It has enough of the Alphabet, Arabian numbers from 0-9, punctuation marks, mathematical notation keys, and I even fit in a variable/universal currency symbol making 48 printable characters, and 2 basic control characters, all in 26 (25) possible values.

It's designed to be printed on punch cards in the least number of columns possible. The benefits of using a ternary system are much higher data density compared to 7/8-bit ASCII, while still being somewhat easy to interpret by a human reader. Since there isn't much media made for ternary computers, I figure it might feasibly store binaries put through some kind of Base25 binary-to-text converter.
On a punch card, data would probably be printed in this form:
1 2 3 4 5 6
201 111 010 221 000 000
210 212 102 011 000 000
010 221 222 210 000 000
102 010 110 011 000 000
221 120 222 012 000 000
Except with whatever the most efficient number of columns and rows, and most importantly groups would be. IBM figured that was 80*12*1, but when the priority's information density at a large, comprehendible and non-volatile size, you've got to find some way to make better use of the room you've got. To interpret it, you go through each of the 6 "groups" here in order, I think a computer should be capable of doing that itself as long as the hardware can handle variable numbers of groups. To work the "2" in there, you could probably puncture just half of a single slot, if you've been finding this difficult to imagine.

Are there any important things I missed? Reasons the idea is dumb and would never work in the first place, bad decisions in the characters I chose, or even reasons it wouldn't really work on a typewriter? I regret that you need to shift before every . with this code in particular, but I don't think there's any more characters there that I can really massacre without having to make people enter more characters than what should be necessary.
As a side-note, Russian ternary computers dealt with "balanced" ternary which used -, 0, and +. Does anybody know why that is? Aside from the notation, is anything mechanically different?

Obviously the next question is, why not go to base-4? You could puncture the top or bottom half of a slot, or even work up to base-5 for a whole slot punctured out. To demonstrate just how much data a base-5 system could store, 3-quintit has 124 (123) values, and 4-quintit does the terrifying number of 629 (628) values. For reference, 8-bit ASCII goes up to 256 (255). While it might be good for storage, it'd probably be annoying for a person to read and manually transcribe if the situation ever came up, regardless of whether it's meant to be read by a man or a machine.
I've been considering whether it'd be worth going that high for that drawback.

By the way, with 4 quadrits, there's 256 values including 0.

Oh, guess I should've used tabs.

Also, not that it really matters, but + and ' were meant to be swapped on the key there. Don't know what happened there.
God, I hate this shitty-ass markup.

>>4
Your confusion between bold and monospace is not the fault of the markup.

>>5
That was just for emphasis. If this board actually had markup for monospace then yeah, I would've used that.

>>6
[code] is monospaced, newfag

retarded. Quartz & marble etching is far better products than soak-able punchcards. I rather memorize the data than to waste material.

>>8 Punch cards can be made out of plastic and metal sheets as well, but obviously that's more expensive.

>>9 Why did you bother responding when you knew you would yield to the correct thought?
Why go centimeter storage, when were are in the nano-space?
I'm still dead serious that I want to see 5D crystal available commodity storage in my lifetime. But I don't see anyone innovating means to produce nano lenses that commodified.
Same thing with optical disc media: who're going to maintain reader drivers to last the lifetimes of their written counterparts?
I keep reading stories of people buying into to the DVD/Blu-ray craze without accessing the lifetime of their lenses, carriages, needles, etc. and how repairable and modifiable they are if at all.

I am super serious, I rather carve on fucking marble that lasts millenia than punch on soluble polymers and oxidizing metals.

Envy Sumerians, their Code of Ur-Nammu exist today and is reinterpretable, while Yeshua has no physical testaments to account their existence. Fuck, even Ea-Nasir marbled their complaints for everyone to resurface!

s/access/asses/
as in assessment: I need to test this shit.

As a side-note, Russian ternary computers dealt with "balanced" ternary which used -, 0, and +. Does anybody know why that is?

It was more efficient. Shame it didn't catch on.

>>10
I see what you're getting at, but stone inscription has two issues as a storage medium.
1. Over the years, it might be worn down, and the inscriptions would become indiscernible.
2. If you do something crazy like emgrave a computer program into it, you're going to have to enter that into the computer by hand.

With the equally, if not more low-precision medium of a punched card though, it could be read within minutes, and, if you wanted, maybe even re-written to marble columns within hours with the appropiate peripheral.

>>12
How is that different from 0, 1, and 2?

>>13
>How is that different from negative flow, no flow, and positive flow
I swear, generation after generation is getting dumber and dumber.

>>13
it can represent negative numbers without storing the sign separately

>>14
This is what happens with too much abstraction. People don't have the slighest clue how hardware functions.

>>14 >>16
That's not what I asked you dumb farts. >>12 stated that -, 0 and + weren't just more semantically accurate than counting to 2, but more EFFICIENT. Assuming there hasn't been some kind of misunderstanding on my part, I want to know why that is. Is it just more intuitive to teach, since 0 usually means 0, rather than -1?

>>15
Yeah, I guess that saves a couple of keypresses.

>>17
Lowering your post when asking, huh.
It's current flow, retard. Shift flow makes addressing simple when doing operations instead of going through a cycle.